readme.txt

linux下的无线宽带驱动

   There are a number of ways of starting the connection. The most usual is to
   issue an ifup command on the ppp interface:
      $ ifup ppp0

   And use ifdown to stop the connection:
      $ ifdown ppp0

   On some distros, it appears that the ifdown command does not properly shut
   the connection down. If you find there are pppd or similar processes
   remaining after an "ifdown ppp0" that should not be there, then your distro
   may have this problem. In this case, try using adsl-start/adsl-stop
   instead.

   The roaring Penguin PPPoE dialler supplies start and stop scripts. To
   start the connection, issue:
      $ adsl-start

   This will start up the first ppp interface configured using adsl-setup.
   If you have more than one ppp interface, you can specify the one you want on
   the adsl-start command line:
      $ adsl-start <path-to>/ifcfg-ppp0

   To disconnect again, you can use the adsl-stop command:
      $ adsl-stop

   or
      $ adsl-stop <path-to>/ifcfg-ppp0


   **Note:
   Some people have found that regardless of the value specified to adsl-setup
   regarding on-demand startup and timeout, adsl-setup seems to configure the
   interface with an excessively short timeout value. See point 5.6 below for
   more details.

   Below is what a successful connection looks like:

   localhost pppd[6801]: pppd 2.4.2 started by root, uid 0
   localhost pppd[6801]: Using interface ppp0
   localhost pppd[6801]: Connect: ppp0 <--> /dev/pts/8
   localhost pppoe[6802]: PPP session is 2
   localhost pppd[6801]: CHAP authentication succeeded
   localhost pppd[6801]: local  IP address XXX.YYY.ZZZ.WWW
   localhost pppd[6801]: remote IP address AAA.BBB.CCC.DDD
   localhost pppd[6801]: primary   DNS address GGG.HHH.III.JJJ
   localhost pppd[6801]: secondary DNS address KKK.LLL.MMM.NNN
   localhost NET: /etc/sysconfig/network-scripts/ifup-post : updated /etc/resolv.conf

   If the startup gets stuck in a fairly tight loop failing to connect, try
   killing the connection attempt, and trying again.


5.  What can go wrong?
    A few things. Here is a summary of the well-known ones, and what to do to
    sort out why:

5.1 Building the drivers

5.1a Error message: /lib/modules/x.y.z/build: No such file or directory
    This indicates that the components required to build modules have not been
    installed/updated.
    
    - If this is the first time you have attempted to build a module,
      then find out what components are required to build modules, and install
      those. (Eg kernel-devel.)

         example fix (check for your distro):
	 $yum install kernel-devel

    - If you have just upgraded your kernel, then you probably need to update
      the components required to build modules as well. (Eg kernel-devel.)

         example fix (check for your distro):
	 $yum update kernel-devel


5.2 Checking the hardware status.
    Confirm the hardware has been recognised.

5.2a PCMCIA
     Use the cardctl command to show what hardware is currently recognised
     as being inserted:

       $ cardctl ident

       or, for the newer pccard systems:

       $ pccardctl ident

     Make sure your iBurst pcmcia card is listed, and compare the manfid
     values to those you entered into the pcmcia config (step 2).

     If the card does not show up when it is inserted, make sure pcmcia is
     configured and started on your machine. Check the logs and dmesg, looking
     for output from pcmcia and the kernel regarding pcmcia.

     If you see the error: "cs: pcmcia_socket0: time out after reset.", and
     cannot fix it in the configuration, try generating an insert event by
     executing the following command (as root):

        $ pccardctl insert

     You may need to repeat this command 2 or 3 times.

     If pcmcia is definitely working, and the card is still not being
     recognised, try testing the hardware to make sure it is not defective.

     If the card shows up, but the manfid differs to that in your config, try
     changing your config to match the card.


5.2b USB
     Use the lsusb command to show what hardware us currently recognised as
     being attached:

       $ lsusb

     Make sure your iBurst modem is listed, and compare the identifiers to
     those you entered into the config (step 2).

     If the modem does not show up when it is attached, make sure usb is
     installed and configured on your machine. Check the logs and dmesg,
     looking for output from usb and the kernel regarding usb. If usb is
     definitely working, and the modem is still not being detected, try
     testing the hardware to make sure it isn't defective.

     If the modem shows up, but the identifiers differ to what you've
     configured, try changing your config to match the modem.


5.3 Confirm the iburst drivers have recognised the hardware.
    The ib-net driver creates an entry in the /proc/driver tree.

    $ cat /proc/driver/iburst
    There is one, single-line entry per recognised device. Each entry shows
    the MAC address of the device.
    Eg: 00:10:E2:04:00:31

    $ ifconfig ib0
    Output looks like (partial):
            ib0   Link encap:Ethernet  HWaddr XX:YY:ZZ:AA:BB:CC
                  BROADCAST NOARP DYNAMIC  MTU:1500  Metric:1
                  RX packets:494 errors:0 dropped:0 overruns:0 frame:0
                  TX packets:9806 errors:0 dropped:0 overruns:0 carrier:0
                  collisions:0 txqueuelen:20
                  RX bytes:4030 (3.9 Kb)  TX bytes:78448 (76.6 Kb)

    Explanation:
        This lists network devices registered to the system;
        those entries beginning with "ib" (or "nas-ib", with debian pppoe) are
        iburst interfaces. The digit(s) at the end of the interface name
        correspond to the entries in /proc/driver/iburst. If the corresponding
        entry in /proc/driver/iburst does not match the HWaddr entry in
        ifconfig, then the drivers have not recognised your device correctly.
        Check your log file or load the module manually to find out why.

    [ibdriver >= 1.3.2]
       As of version 1.3.2, ibdriver publishes information to the linux
       wireless extensions.

       $ iwconfig ib0
       Output looks like:
	  ib0    HC-SDMA (iBurst)  Bit Rate:0.112 kb/s
		 Link Quality:100/100  Signal level:100/100  Noise level:0/100
		 Rx invalid nwid:0  Rx invalid crypt:0  Rx invalid frag:0
		 Tx excessive retries:0  Invalid misc:0   Missed beacon:0
      
    [/ibdriver >= 1.3.2]

5.3.1 Confirm modules have loaded correctly.
      **Note: The names of the modules on disk are ib-* whereas the name they
      report to the kernel are ib_*.

      Insert your device, and then check to ensure the correct modules have
      loaded:

        $ lsmod | grep ib_

      If the ib-net driver has not loaded, then dependent drivers including
      ib-pcmcia and ib_usb will not load correctly, and will report missing
      symbols (from the ib-net driver). Make sure the ib-net driver has loaded
      correctly before trying to diagnose errors with dependent drivers.

      **Note: misspelled driver options will typically stop a driver from
      loading. If a driver fails to load, and no other reason can be
      determined, try commenting out all "options" lines for the driver, and
      loading the driver again. If this works, check each value in the
      "options" line for typing mistakes.

5.3.1a PCMCIA:
       You should see ib-net and ib-pcmcia are loaded.
       If not, load them manually, to find out why.

       $ modprobe ib-net
       $ modprobe ib-pcmcia

5.3.1b USB
       You should see ib-net and ib-usb are loaded.
       If not, load them manually, to find out why.

       $ modprobe ib-net
       $ modprobe ib-usb


5.3.2 /proc/driver/iburst does not exist.
      The ib-net module did not load. Check your log file, or load the module
      manually to find out why.

5.3.3 ib0 (or nas-ib0) interface is not found.
      The ib-net module has not created the interface properly.
      Possible causes include incorrect configuration, or module(s) failing to
      load.

      - Check your configuration to ensure the interface name is correct and
        consistent

      - Check the log file or load the modules manually to see if there are any
        errors (see 5.3 above)


5.4 Error message: "unrecognised option" when loading the ib-pcmcia module.
    Check the spelling of the CORE_OPTS and/or pcmcia_core options.


5.5 Error message:
    "cs: socket <abc> timed out during reset.  Try increasing setup_delay."

    This error is usually accompanied by both lights on the card going off
    1-2 seconds after insertion.

    The pcmcia core is timing out before the iburst card has initialised.
    This can be fixed by passing some options to pcmcia_core on startup.
    The usual problem is misspelled options, or options that are not being
    passed to the pcmcia_core module. For example, Fedora Core doesn't always
    pass CORE_OPTS to the pcmcia module.

    If you added options to CORE_OPTS (step 2 above), check them for
    typos, and make sure the pcmcia module has been unloaded and reloaded. If
    this error continues, add the following line to your modules config file:

    "options pcmcia_core unreset_delay=100 unreset_check=20 unreset_limit=100"

    The option that specifically fixes this problem is the "unreset_limit"
    option. (That's correct, adjusting "setup_delay" doesn't actually fix the
    problem).

    If you continue to get this error after doing all the above, then the
    pcmcia module is probably not being reloaded, so try a reboot.

5.5.1 Error message:
    "cs: pcmcia_socket0: time out after reset."

    This error is usually accompanied by both lights on the card going off
    1-2 seconds after insertion.

    If this is on a kernel >= 2.6.17, we have not yet found a configuration
    that fixes this automatically. The current manual workaround is to 'insert'
    the card through software by executing the following command (as root):

       $ pccardctl insert

    You may have to repeat the command 2 or 3 times before the card initialises
    correctly.

5.6 Error message:
    "Inactivity timeout... something wicked happened on session XX"
    The connection is being closed due to a timeout. You can try adjusting the
    PPPOE_TIMEOUT parameter in the PPP config file. Eg:
    edit /etc/sysconfig/network-scripts/ifcfg-ppp0 (may be somewhere else in
    your distro - you may also try using the GUI admin tools to adjust the
    parameters of ppp0).

    Change PPPOE_TIMEOUT="xx" to some larger value, eg PPPOE_TIMEOUT="3600"
    should increase the timout period to 1 hour. Setting PPPOE_TIMEOUT to a
    value of zero (eg PPPOE_TIMEOUT=0) should disable the timeout altogether.


5.7 Checking the logs.
    If all is working as it should, there are no spurious errors in the log.
    All errors should be treated as a sign that something is wrong.

5.7.1 Error message:
    localhost wait_for_sysfs[xyz]: either wait_for_sysfs (udev abc) needs an
    update to handle the device '/class/net/nas-ib0' properly
    (no device symlink) or the sysfs-support of your device's driver needs to
    be fixed, please report to <[e-mail]>

    Updating udev normally fixes this error. Beware of simply downloading and
    building the very latest udev - it is probably better to get a version
    built for your release. For example, at the time of this writing, the
    latest udev version is 057, and Fedora Core 3 ships with 039. Updating
    manually to version 057 resulted in errors that made X unusable, whereas
    installing the Fedora update to 039-10 fixed the problem. Using an update
    program is probably a very good approach, eg: "$ yum update udev".


5.7.2 Error message(s)
    localhost cardmgr[xxx]: socket 0: ArrayComm ut02
    localhost kernel: ib-pcmcia: Unknown parameter `'

    Some recent edit has probably left one or more spaces at the end of an
    "options" lines in your file. Edit yout modules config file and delete any
    trailing spaces from all lines.

    Possibly, you've removed a value from an "options" line in the modules
    configuration file, or perhaps had some other module installed since you
    installed the iburst driver, possibly by kudzu, or some other automated
    tool which has added one or more entries to the module configuration
    (/etc/modprobe.conf or /etc/modules, etc).

6. Cleaning up
Once everything is working smoothly, you should disable any debug output.

6.1 Edit the modules config file (eg /etc/modprobe.conf) and remove any
    "debug=" string from the "options ib-pcmcia" and/or "options ib-usb"
    line(s) (if it is present).


Appendices

A1. Suspend/Resume
  The pcmcia driver (ib-pcmcia) correctly suspends and resumes. However, the
  ib0 network interface is not correctly reinitialising. So currently, if
  you wish to suspend the machine, you should disconnect the pppoe connection,
  and eject the iburst card. Once you have resumed, re-insert the card and
  re-connect.

  It is expected that a future release will support suspend/resume without
  needing to eject and reinsert the card.


-------------------------------------------------------------------------------
* "ArrayComm" is a registered trademark of ArrayComm inc.
* "Iburst" is a tradmark of ArrayComm inc.
* "PBA" and "Personal Broadband Australia" are trademarks of Personal Broadband
   Australia.
-------------------------------------------------------------------------------